My use-case is determining the sequence in which to run object creation files. Each object (SP, function, type etc.) is in its own file. Each table's file include the PK, FKs, indexes and triggers so they don't need separate files.
This sorting is then input to a devops-style DB creation Powershell which submits each object file in dependency sequence ensuring no errors.
There's a parallel one for populating reference data. It's the same but looking only at tables' FKs.